// ◉ Multimodal & Vision
wav2vec 2.0
Contrastive self-supervised pre-training on raw audio; strong ASR from little labelled data.
wav2vec 2.0 自监督语音表征 2020 Intermediate core
IN PLAIN WORDS
A conv encoder turns the waveform into frames, some are masked, and a Transformer must pick the true quantized representation for each masked slot from distractors (contrastive learning). Pre-training needs only unlabelled audio; fine-tuning with a little transcribed speech and CTC then gives strong recognition – the paper reports usable results with just 10 minutes of labels.
Used in: Low-resource speech recognition and speech features (a common ASR model on Hugging Face)
TIMELINE · prerequisites → this → what builds on it
01Original paper
wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations
Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, et al. · 2020
arXiv ↗ ✓ Verified · arXiv API
02Where it sits
Learn the left column first; the right column builds on this one.
Then you can learn
- Nothing in the atlas builds on it yet
Related:WhisperConnectionist Temporal Classification (CTC)
Left: prerequisites · right: builds on this · dashed: related. Full map →
03Learning resources
Official docs, explainers, courses and reference code. All links checked.
Official docs
Reference code
Courses
04Where the course teaches it
For AI / data
- Verification: verified, checked 2026-10-22.
- JSON: /api/learn/algorithms.json (id:
wav2vec2)